Epidemiology/Health Services Research

O R I G I N A L A R T I C L E

Breakfast Frequency and Development

of Metabolic Risk
ANDREW O. ODEGAARD, PHD1 LINDA VAN HORN, PHD2 of weight gain/obesity, metabolic syn-
DAVID R. JACOBS JR., PHD1 DAVID S. LUDWIG, MD, PHD3 drome, hypertension, and type 2 diabetes
LYN M. STEFFEN, PHD1 MARK A. PEREIRA, PHD1 (8–14). An important interpretative con-
sideration is that breakfast intake fre-
quency has generally been dichotomized
OBJECTIVEdThe relation of breakfast intake frequency to metabolic health is not well stud- into an all-or-none proposition, although
ied. The aim of this study was to examine breakfast intake frequency with incidence of metabolic consistent evidence across cultures and
conditions.
populations suggests there is a range of
RESEARCH DESIGN AND METHODSdWe performed an analysis of 3,598 partici- breakfast intake frequencies (8,15,16).
pants from the community-based Coronary Artery Risk Development in Young Adults (CARDIA) Therefore, we examined the associa-
study who were free of diabetes in the year 7 examination when breakfast and dietary habits were tion of a range of breakfast intake fre-
assessed (1992–1993) and participated in at least one of the five subsequent follow-up exami- quencies with risk of an array of incident
nations over 18 years. metabolic outcomes over 18 years in the
RESULTSdRelative to those with infrequent breakfast consumption (0–3 days/week), partic-
Coronary Artery Risk Development in
ipants who reported eating breakfast daily gained 1.9 kg less weight over 18 years (P = 0.001). In a Young Adults (CARDIA) study, a multi-
Cox regression analysis, there was a stepwise decrease in risk across conditions in frequent center, population-based, prospective
breakfast consumers (4–6 days/week) and daily consumers. The results for incidence of abdom- study of cardiovascular risk evolution in
inal obesity, obesity, metabolic syndrome, and hypertension remained significant after adjust- young black and white adult men and
ment for baseline measures of adiposity (waist circumference or BMI) in daily breakfast women in the U.S. We hypothesized that
consumers. Hazard ratios (HRs) and 95% CIs for daily breakfast consumption were as follows: breakfast eating would show a graded in-
abdominal obesity HR 0.78 (95% CI 0.66–0.91), obesity 0.80 (0.67–0.96), metabolic syndrome verse relationship with incident metabolic
0.82 (0.69–0.98), and hypertension 0.84 (0.72–0.99). For type 2 diabetes, the corresponding conditions, partially explained by quality of
estimate was 0.81 (0.63–1.05), with a significant stepwise inverse association in black men and the overall diet.
white men and women but no association in black women. There was no evidence of differential
results for high versus low overall dietary quality.
RESEARCH DESIGN AND
CONCLUSIONSdDaily breakfast intake is strongly associated with reduced risk of a spec- METHODS
trum of metabolic conditions.
Study and data collection
Diabetes Care 36:3100–3106, 2013 The CARDIA study is a multicenter, longi-
tudinal investigation of the evolution of

T
here is a historical precedent for range of study designs generally demon- ischemic heart disease risk starting in
breakfast intake being linked with strating that both the timing (breaking young adulthood (17). The study began
health. The earliest documented of a fasting state) and content of breakfast in 1985–1986 with 5,115 black and white
claims were from fledgling ready-to-eat may be important for health, especially adults 18–30 years of age from four metro-
cereal companies in the 1800s and from a metabolic health, via interrelated mecha- politan areas (Birmingham, AL; Chicago,
pork producer in the 1920s touting phy- nisms involving metabolism and appetite IL; Minneapolis, MN; and Oakland, CA).
sician recommendations to eat a hearty (6). On the basis of the existing data, the Study participants were sampled to obtain
breakfast of bacon and eggs (1,2). Midway 2010 U.S. dietary guidelines were the first roughly equal numbers of blacks (51.5%)
through the 1900s, small studies finding to include a specific recommendation for and whites (48.5%), men (45.5%) and
potential health benefits of breakfast be- breakfast intake (7). women (54.5%), 18–24 years of age
gan appearing in the scientific literature Several studies provide prospective (44.9%) and 25–30 years of age (55.1%),
(3–5). In the time since, investigation evidence directly linking either the be- and with a high school education or less
on a number of aspects related to break- havior of eating breakfast or consumption (39.7%) and with more than a high school
fast intake and health has occurred with a of typical breakfast foods with lower risk education (60.3%). Participants were con-
tacted by telephone every year and exam-
c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c c ined in person at baseline and 2, 5, 7, 10,
From the 1Division of Epidemiology and Community Health, School of Public Health, University of Minne- 15, 20, and 25 years after baseline. A ma-
sota, Minneapolis, Minnesota; the 2Department of Preventive Medicine, Northwestern University Medical jority of the group was examined at each of
School, Chicago, Illinois; and the 3New Balance Foundation Obesity Prevention Center, Boston Children’s
Hospital, Boston, Massachusetts.
the follow-up examinations (91, 86, 81, 79,
Corresponding author: Andrew Odegaard, odeg0025@umn.edu. 74, 72, and 72% of survivors, respectively).
Received 6 February 2013 and accepted 17 April 2013. At each clinical examination, participants
DOI: 10.2337/dc13-0316 were asked to present fasting in the morning.
This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/suppl/doi:10 Tobacco use, strenuous physical activity,
.2337/dc13-0316/-/DC1.
© 2013 by the American Diabetes Association. Readers may use this article as long as the work is properly and intake of caffeine, food, and alcohol
cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/ were proscribed. The examinations followed
licenses/by-nc-nd/3.0/ for details. standardized protocols harmonized over

3100 DIABETES CARE, VOLUME 36, OCTOBER 2013 care.diabetesjournals.org


time and included measurements of blood
pressure (BP), anthropometrics, and phle-
botomy and structured questionnaires on
sociodemographics, medical and family
history, psychosocial characteristics, and
diet, among others. The CARDIA study
was approved by the institutional review
board of each participating institution,
and signed informed consent was obtained
at each examination.
During each clinic exam, blood was
drawn from an antecubital vein, and after
serum separation, aliquots were stored at
2708C until shipped on dry ice to a cen-
tral laboratory. Procedures followed in
the collection and storage of plasma sam-
ples, laboratory quality-control proce-
dures, and methodology for analysis of
plasma triglycerides, HDL cholesterol,
LDL cholesterol, and total cholesterol are
described elsewhere (18). Serum glucose
was measured at year 0 using the hexoki-
nase ultraviolet method by American Bio-
Science Laboratories (Van Nuys, CA) and at
subsequent examinations using hexoki-
nase coupled to glucose-6-phosphate de-
hydrogenase by Linco Research (St. Louis,
MO).
BP was measured three times at 1-min
intervals. At the baseline through year 15
follow-up exams, BP was measured using
the Hawksley (Lancing, Sussex, U.K.)
random-zero sphygmomanometer; the first
and fifth phase Korotkoff sounds were
recorded (17). At the year 20 and 25 exams,
BP was measured with an automated
sphygmomanometer (Omron HEM907XL
oscillometer; Omron, Schaumburg, IL).
The protocol specified the appropriate
cuff size (small, medium, large, or extra-
large) based on the upper arm circumfer-
ence, which was measured by the BP
technician at the midpoint between the ac-
romion and the olecranon. Omron values
were recalibrated to corresponding random
zero values based on a study of both mea-
surement techniques in 903 participants at
year 20, as estimated random zero systolic
value = 3.74 + 0.96 3 Omron systolic
value, and estimated random zero dia-
stolic value = 1.30 + 0.97 3 Omron dia-
stolic value (19).
Anthropometry (height, weight, and
waist circumference) was measured at
each exam. Body weight was measured
to the nearest 0.2 kg using a calibrated
balance beam scale in participants wear-
ing light clothing. Height (without shoes)
was measured to the nearest 0.5 cm
using a vertical ruler and waist circum-
ference to the nearest 0.5 cm at the
minimal abdominal girth (20). BMI was
care.diabetesjournals.org
computed as weight in kilograms divided
by squared height in meters.
Diet was assessed at years 0, 7, and 20
by using an interviewer-administered
CARDIA diet history questionnaire (21).
Interviewers asked open-ended questions
about dietary consumption in the past
month within 100 food categories that
referenced 1,609 separate food items. Ad-
ditionally, visits per week to fast food
restaurants were queried at each examina-
tion, and frequency of breakfast, lunch,
dinner, and morning, afternoon, and
evening snacks (days/week) was queried
at years 7 and 20. An a priori dietary qual-
ity score based on overall dietary intake
was included as a covariate (16). In brief,
46 food groups considered beneficial or
adverse with respect to health effects were
categorized by increasing consumption
level with scores of 0–4 (for 20 food
groups considered beneficial), 4–0 (for
13 food groups considered adverse), or
0 (for 13 food groups considered neutral).
The a priori dietary quality score was the
sum of category scores, with a theoretical
maximum of 132. Based on prior find-
ings, it was assumed that a higher a priori
dietary pattern score indicated better diet
quality (22–24). The CARDIA diet history
has been shown to be a valid and reliable
instrument. Nutrient and energy esti-
mates had larger variability among blacks
than among whites (21,25). For other co-
variates, standard questionnaires were
used to obtain self-reported demographic
and behavioral information. Sex, race,
date of birth, education, and cigarette
smoking were ascertained by a struc-
tured interview or self-administered
questionnaire at each examination. A
physical activity score was derived
from the CARDIA physical activity his-
tory, which is a simplified version of the
Minnesota Leisure Time Physical Activ-
ity Questionnaire (26).
Assessment of outcomes
Year 7 (1992–1993) serves as the baseline
for this study. Incident outcomes were
identified at exam years 10–25. Obesity
was defined as BMI $30 kg/m2 and ab-
dominal obesity as waist circumference
.88 cm for women or .102 cm for
men. Hypertension was defined as sys-
tolic BP $140 mmHg, diastolic BP $90
mmHg, or self-reported use of antihyper-
tensive medication. The metabolic syn-
drome was defined using the National
Cholesterol Education Program (NCEP)
Adult Treatment Panel (ATP) III criteria
(27) as the presence of three or more of
DIABETES CARE, VOLUME 36, OCTOBER 2013
Odegaard and Associates
the following five conditions: 1) abdomi-
nal obesity, 2) fasting triglycerides $150
mg/dL, 3) HDL cholesterol ,40 mg/dL in
men and ,50 mg/dL in women, 4) BP
$130 mmHg systolic or $85 mmHg di-
astolic or use of antihypertensive medica-
tions, and 5) fasting glucose $100 mg/dL
or use of diabetes medications. Incidence
of type 2 diabetes was defined as use of
diabetes medication (assessed at every
visit), a fasting blood glucose level of
$6.99 mmol/L (126 mg/dL) (measured
at years 10–25), 2 h postchallenge glucose
$11.1 mmol/L (200 mg/dL) (performed
at the year 10, 20, and 25 exams), and/or
an HbA 1c $6.5% (48 mmol/mol) (as-
sessed at the year 20 and 25 visits).
Statistical analysis
From the total sample of 5,115, we
excluded 1,029 who did not participate
in the year 7 clinical exam, a further 171
who did not participate in any clinical
exam in years 10–25, 60 with diabetes at
year 7, 215 without dietary data or with
reported energy intakes not in the range
of 800–8,000 kcal/day for men and
600–6,000 kcal/day for women, and 42
missing other data (alcohol, smoking, or
physical activity) for an analytic sample of
3,598. For analyses on outcomes other
than type 2 diabetes, participants preva-
lent with the condition at year 7 were also
excluded and total n is provided in tables.
Breakfast intake frequency catego-
ries were created that allowed for logical
cut points with a sufficient number of
subjects. Baseline (year 7) characteristics
were calculated across breakfast intake
frequency categories reported at this
exam. Multivariable least squares–
adjusted means from general linear mod-
els (SAS Proc GLM) were used to estimate
weight gain in kilograms and increase in
waist circumference in centimeters by
breakfast intake frequency categories.
The models adjust for age, study center,
race, sex, education (years), cigarette
smoking (current, former, or never), phys-
ical activity (units/week), alcohol con-
sumption (mL/day), fast food restaurant
use (visits/week), overall dietary quality
score, frequency of lunch/dinner and
morning/afternoon/evening snacks (days/
week), total energy intake (kcal), and
weight or waist circumference and height
at year 7, respectively. All participants
were free of diabetes at each time point
in the analysis of weight and waist circum-
ference changes.
Proportional hazards (Cox) regression
(SAS Proc PHREG) was used to examine
3101
Breakfast and metabolic risk

the association between breakfast intake
frequency categories and incidence of met-
abolic conditions. We estimated the hazard
ratio (HR) and corresponding 95% CI.
Time to event was calculated from the
date of the baseline examination (year 7) to
the date of the first follow-up examination
meeting the criteria for the incident out-
come (cases) or to the date of the last
CARDIA examination for each participant
without the incident outcome (censored).
A tiered modeling approach was applied
for all outcomes. The main model included
age (years), study center, race, sex, educa-
tion (years), cigarette smoking (current,
former, or never), physical activity (units/
week), alcohol consumption (mL/day), fast
food restaurant use (visits/week), dietary
quality score, frequency of lunch/dinner
and morning/afternoon/evening snacks
(days/week), and total energy intake
(kcal). Depending on the outcome, either
waist circumference (cm) or BMI (kg/m2)
from year 7 was included in a second
model, and, for hypertension, a third
model further adjusted for systolic BP at
year 7 in analyses examining potential me-
diators. There was no evidence that pro-
portional hazards assumptions were
violated for any of the outcomes as indi-
cated by the lack of significant interaction
between the breakfast intake frequency var-
iable and time in the models. Tests for trend
were performed by assigning the median
value of intake frequency to the category
and entering this as a continuous ordinal
variable into the models. Effect modifica-
tion of the associations was considered by
level of the dietary quality index, BMI, race,
and sex. All analyses were conducted with
SAS statistical software version 9.2 (SAS In-
stitute, Cary, NC).
RESULTSdBased upon the year 7 data,
43.2% of participants reported infrequent
breakfast intake (0–3 days/week), 21.7%
reported eating breakfast frequently (4–6
days/week), and 35.1% of participants re-
ported eating breakfast daily (7 days/week)
(Table 1). With higher levels of breakfast
intake, a greater proportion of participants
were white and female and were on average
more educated, consumed less alcohol, did
not currently smoke, were more physically
active, had a lower BMI, visited fast food
restaurants less frequently, and had a
higher dietary quality score.
Over 18 years of follow-up, there was a
significant mean weight gain in all par-
ticipants free of diabetes throughout the
study; yet, frequent (4–6 days/week) and
daily breakfast consumers gained less
Table 1dParticipant characteristics according to breakfast frequency (days per week):
CARDIA year 7, 1992–1993
Breakfast frequency (days/week)
0–3
n 1,556
Age (years) 31.8 (3.8)
Race (% black) 59.9
Sex (% female) 53.3
Education (years) 14.1 (2.3)
Alcohol (mL/day) 12.4 (22.4)
Smoking (% current) 34.9
Smoking (% former) 14.4
Physical activityx 319.7 (268.2)
BMI (kg/m2) 27.9 (6.5)
Weight (kg) 81.3 (19.3)
Waist (cm) 86.5 (14.2)
Hypertension (%) 5.3
Fast food (visits/week)U 3.6 (4.9)
Dietary quality score 64.2 (11.6)
Energy intake (kcal/day) 2,765 (1,255)
Data are unadjusted mean (SD) for all characteristics unless noted as %. xPhysical activity, exercise units.
UFast food, fast food restaurant outlet (visits/week).
weight relative to non- or infrequent break-
fast consumers (Supplementary Fig. 1A
and B). Specifically, participants reporting
daily breakfast intake gained 1.91 kg less
than those reporting infrequent intake (0–3
days/week) (P = 0.001) over 18 years after
full adjustment for demographic, life-
style, dietary habits, and baseline weight.
A similar trend was observed for waist
circumference.
Across all metabolic outcomes, there
was a stepwise decrease in crude incidence
rate, and the incidence rate was nearly
halved in daily breakfast consumers relative
to those who were infrequent breakfast
consumers (0–3 days/week) (Table 2).
These graded associations were evident
in the main Cox regression model adjust-
ing for demographics, lifestyle covariates,
and dietary habits. Relative to infrequent
intake of breakfast, frequent breakfast in-
take (4–6 days/week) and daily breakfast
intake were each significantly associated
with a decreased risk of abdominal obe-
sity, obesity, metabolic syndrome, hyper-
tension, and type 2 diabetes in a ranked
manner. After adjustment for baseline
measures of adiposity (waist circumfer-
ence [cm] or BMI [kg/m2]), the associa-
tions were attenuated but a significant
inverse association persisted between
daily breakfast intake and abdominal
obesity, obesity, metabolic syndrome, and
hypertension.
The estimates for incident type 2 di-
abetes were mediated upon adjustment for
4–6 7
779 1,263
32.0 (3.5) 32.4 (3.4)
47.8 28.0
53.7 59.8
14.8 (2.5) 15.3 (2.5)
10.2 (18.4) 8.7 (19.9)
23.0 15.8
14.5 20.2
338.1 (262.1) 354.5 (266.9)
26.7 (5.9) 25.1 (5.2)
78.3 (19.1) 73.2 (17.0)
83.8 (13.5) 80.0 (12.5)
5.1 3.3
3.5 (3.8) 2.3 (3.0)
66.6 (11.7) 70.6 (12.4)
2,762 (1,222) 2,743 (1,160)
BMI in the whole-population HR and 95%
CI for frequent breakfast intake (HR 0.82
[95% CI 0.63–1.07]) and daily intake (0.81
[0.63–1.05]). However, there was evidence
that the results for type 2 diabetes dif-
fered in black women from those for the
rest of the study sample (Table 3). In
black women, breakfast intake frequency
was not associated with incident type 2
diabetes, whereas the results were consis-
tent and strongly inversely associated in
black men and white men and women even
after adjustment for BMI. Black women had
the highest rate of incident diabetes in the
cohort and greatest mean level of BMI in
year 7 (29.0 kg/m2) relative to the rest of the
study population (black men, 27.0 kg/m2;
white men, 26.0 kg/m2; white women,
24.9 kg/m2). Adjustment for hypertensive
status and medication did not materially
alter the results for incident type 2 diabetes.
There was no evidence that the results
varied for any of the other outcomes by
race, sex, or BMI at baseline. There was no
evidence that adjustment for family history
of type 2 diabetes or hypertension mate-
rially altered any of the results. The fre-
quency of lunch, dinner, or snacks was not
associated with any of the outcomes. Of
note, we also examined the association be-
tween breakfast intake frequency and fu-
ture dyslipidemia (low HDL cholesterol and
elevated triglycerides per ATP III criteria).
There was an inverse, but nonsignificant,
association with greater breakfast intake
(data not presented).

3102 DIABETES CARE, VOLUME 36, OCTOBER 2013 care.diabetesjournals.org

 Odegaard and Associates

Table 2dHR and 95% CI of metabolic outcomes according to breakfast frequency: CARDIA We also performed a sensitivity analy-
years 7–25 (1992–1993 to 2010–2011) sis with metabolic syndrome as the out-
come exploring the statistical effect of
Breakfast frequency (days/week) adjustment for common breakfast foods
in the study (timing of consumption was
0–3 4–6 7 P trend not asked). The whole grain breakfast
Abdominal obesity cereal food group was the only breakfast-
n cases/n 470/1,138 202/610 300/1,101 oriented food to materially alter the point
Incidence rate 23.4 17.4 13.4 estimates in model 2 with metabolic syn-
Model 1: HR (95% CI) 1.0 0.75 (0.63–0.89) 0.60 (0.51–0.71) ,0.0001 drome as the outcome (HR 0.86 [95% CI
Model 2a: HR (95% CI) 1.0 0.84 (0.70–0.99) 0.78 (0.66–0.91) 0.001 0.72–1.02]) for daily breakfast relative to
Obesity infrequent breakfast. The refined grain ce-
n cases/n 392/1,064 165/579 226/1,068 reals, eggs, sausage/processed meats, fried
Incidence rate 20.3 14.6 10.1 potatoes, and donuts/pastries/cakes food
Model 1: HR (95% CI) 1.0 0.75 (0.62–0.90) 0.57 (0.47–0.68) ,0.0001 groups did not materially alter the point
Model 2b: HR (95% CI) 1.0 0.85 (0.71–1.03) 0.80 (0.67–0.96) 0.011 estimates.
Metabolic syndrome Of note, breakfast frequency at year 7
n cases/n 465/1,375 189/714 246/1,193 displayed an r = 0.46 (P , 0.0001) cor-
Incidence rate 18.1 13.3 9.8 relation with breakfast frequency at year
Model 1: HR (95% CI) 1.0 0.79 (0.66–0.94) 0.63 (0.54–0.75) ,0.0001 20. Accounting for the year 20 breakfast
Model 2b: HR (95% CI) 1.0 0.89 (0.75–1.06) 0.82 (0.69–0.98) 0.02 data significantly depleted the analytic
Hypertension sample (;42–60% of sample depending
n cases/n 525/1,399 209/714 269/1,186 on outcome) since any prospective exam-
Incidence rate 20.3 14.9 10.8 ination between year 20 and 25 exams re-
Model 1: HR (95% CI) 1.0 0.84 (0.72–0.99) 0.74 (0.63–0.86) 0.0001 quired data from year 7 and 20, no history
Model 2b: HR (95% CI) 1.0 0.90 (0.76–1.06) 0.84 (0.72–0.99) 0.027 of the respective metabolic condition at
Model 3: HR (95% CI) 1.0 0.96 (0.82–1.13) 0.86 (0.73–1.00) 0.069 year 20, and attendance at the year 25
Type 2 diabetes exam. In the sensitivity analyses account-
n cases/n 224/1,556 78/779 95/1,263 ing for the average breakfast intake over
Incidence rate 6.7 4.5 3.3 time in this subgroup (years 7 and 20), the
Model 1: HR (95% CI) 1.0 0.76 (0.58–0.98) 0.66 (0.51–0.86) ,0.001 results did not materially differ from the
Model 2b: HR (95% CI) 1.0 0.82 (0.63–1.07) 0.81 (0.63–1.05) 0.07 results presented using the year 7 data.
Therefore, we solely present the year 7 data
Model 1 includes age, study center, race, sex, education, cigarette smoking, physical activity, alcohol con- for simplicity.
sumption, fast food restaurant use, dietary quality score, frequency of lunch/dinner and morning/afternoon/
evening snacks, and total energy intake. Model 2a includes model 1 + waist circumference (cm) at baseline
(year 7). Model 2b includes model 1 + BMI (kg/m2) at baseline (year 7). Model 3 includes model 2b + systolic CONCLUSIONSdIn black and white
BP at baseline (year 7). Abdominal obesity, waist circumference .88 cm women and .102 cm men; obesity, young adult men and women, frequent
BMI $30 kg/m2; metabolic syndrome, defined according to ATP III guidelines; hypertension, systolic BP (4–6 days/week) and daily (7 days/week)
$140 mmHg, diastolic BP $90 mmHg, or self-reported use of antihypertensive medication; incidence rate, breakfast consumption was associated
incidence per 1,000 person-years; type 2 diabetes, use of diabetes medication (assessed at every visit),
a fasting blood glucose level of $6.99 mmol/L (126 mg/dL) (measured at years 10–25), 2 h postchallenge with a decreased risk of developing ab-
glucose $11.1 mmol/L (200 mg/dL) (performed at the year 10, 20, and 25 exams), and/or an HbA1c $6.5% dominal obesity, obesity, metabolic syn-
(48 mmol/mol) (assessed at the year 20 and 25 visits). drome, hypertension, and type 2 diabetes
over 18 years of follow-up relative to their
peers with infrequent breakfast consump-
We hypothesized that the association dietary quality, there was a stepwise de- tion (0–3 days/week). These findings re-
between breakfast intake frequency and crease in incidence rate of metabolic syn- mained significant for daily breakfast
metabolic risk may vary by the quality of drome with greater breakfast intake intake for all outcomes except type 2 dia-
the overall dietary pattern, i.e., any associ- frequency. The highest incident rates of betes after accounting for baseline mea-
ation may be limited to those with higher metabolic syndrome were observed in in- sures of adiposity. However, the inverse
relative diet quality. However, we found no frequent breakfast consumers (0–3 days/ relationship between greater breakfast fre-
evidence in formal tests for interaction or week) in the bottom half of overall dietary quency and type 2 diabetes risk remained
stratified analyses that the relationship quality, whereas the lowest incident rates independent of BMI in black men and
between breakfast intake frequency and of metabolic syndrome were observed in white men and women, whereas in black
metabolic risk was differential by overall the daily breakfast eaters in the top half of women, there was no association between
dietary quality. We present results from the overall dietary quality. In the main strati- breakfast intake and type 2 diabetes inci-
analysis with metabolic syndrome (Table 4) fied Cox regression model (model 1), there dence. Of note, counter to our hypothesis,
as they are typical of findings for the out- was a graded inverse association with inci- the results were not explained by the over-
comes examined. We ranked the dietary dent metabolic syndrome with more fre- all quality of the dietary pattern.
quality score into quartiles, with the lowest quent breakfast intake across overall Prospective research examining a range
representing the poorest overall quality and dietary quality; however, the results were of breakfast intake frequencies with metabolic
the highest representing a theorized best only suggestively significant in the lowest outcomes is limited. In the only other study
overall dietary pattern. Across quartiles of quartile of diet quality. to examine a range of breakfast intakes,

care.diabetesjournals.org DIABETES CARE, VOLUME 36, OCTOBER 2013 3103

Breakfast and metabolic risk

Table 3dHR and 95% CI of type 2 diabetes according to breakfast frequency, stratified consumers (yes vs. no) had a lower risk
results: CARDIA years 7–25 (1992–1993 to 2010–2011) of a 5-kg weight gain over 10 years (9).
In a tangentially related study, greater in-
Breakfast frequency (days/week) take frequency of both refined and whole
grain ready-to-eat cereals was associated
0–3 4–6 7 P trend in a dose-dependent manner with lower
Black men and white men mean weight gain and lower risk of be-
and women coming overweight (BMI 25 kg/m2) (10).
n cases/n 133/998 47/574 62/1,048 In a similar examination of Physicians’
Incidence rate 6.2 3.7 2.6 Health Study data, there was an inverse
Model 1: HR (95% CI) 1.0 0.64 (0.45–0.89) 0.54 (0.39–0.75) ,0.0001 association between greater intake of ce-
Model 2: HR (95% CI) 1.0 0.67 (0.48–0.93) 0.67 (0.49–0.92) 0.005 real and risk of developing hypertension,
Black women although with greater limits in the interpre-
n cases/n 91/558 31/205 33/215 tation due to the dietary assessment (12).
Incidence rate 7.6 7.1 7.1 In a study of Australian children who were
Model 1a: HR (95% CI) 1.0 0.97 (0.64–1.47) 1.00 (0.66–1.50) NS followed up as young adults, those who
Model 2: HR (95% CI) 1.0 1.06 (0.70–1.61) 1.17 (0.78–1.76) NS reported yes at both childhood and young
adulthood at a dichotomous assessment of
Model 1 includes age, study center, race, sex, education, cigarette smoking, physical activity, alcohol con- breakfast consumption (yes vs. no) had
sumption, fast food restaurant use, dietary quality score, frequency of lunch/dinner and morning/afternoon/
evening snacks, and total energy intake. Model 1a includes model 1 but no inclusion of race or sex. Model 2
lower levels of clinical cardiometabolic
includes model 1 + BMI (kg/m2) at baseline (year 7). Incidence rate, incidence per 1,000 person-years; risk factors relative to those who skipped
type 2 diabetes, use of diabetes medication (assessed at every visit), a fasting blood glucose level of $6.99 mmol/L at different life-course points (14).
(126 mg/dL) (measured at years 10–25), 2 h postchallenge glucose $11.1 mmol/L (200 mg/dL) (performed at Two different studies have examined
the year 10, 20, 25 exams), and/or an HbA1c $6.5% (48 mmol/mol) (assessed at the year 20 and 25 visits). aspects related to breakfast in relation to
type 2 diabetes. In the Health Professio-
there was a gradient of BMI change in ing the greatest increase and daily eaters nals Study, men who did not eat breakfast
adolescents across categories of breakfast experiencing the smallest increase (8). In (yes vs. no) were at an increased risk of
frequency, with never eaters experienc- the Health Professionals Study, breakfast developing type 2 diabetes, and those
who had a high Western dietary pattern
Table 4dHR and 95% CI of metabolic syndrome according to breakfast frequency by score and did not eat breakfast experienced
overall dietary quality: CARDIA years 7–25 (1992–1993 to 2010–2011) an even greater risk of incident type 2
diabetes (13). The Physicians’ Health Study
also found that more frequent intake of
Breakfast frequency (days/week) ready-to-eat cereal, especially whole grain
Diet quality score 0–3 4–6 7 P trend cereal, was inversely associated with risk of
incident type 2 diabetes (11). Two other
Quartile 1 (low) studies have linked aspects of breakfast in-
n cases/n 152/447 60/182 50/193 take with reduced risk of mortality during
Incidence rate 18.7 17.8 13.3 their follow-up periods (28,29).
Model 1: HR (95% CI) 1.00 0.94 (0.69–1.27) 0.75 (0.54–1.04) 0.10 In summary, our study and these other
Model 2: HR (95% CI) 1.00 0.97 (0.69–1.34) 0.92 (0.67–1.25) 0.68 studies all suggest that breakfast intake, or
Quartile 2 frequent consumption of foods associated
n cases/n 123/346 48/175 50/229 with breakfast intake, is important for
Incidence rate 19.2 14.0 10.8 metabolic health. CARDIA provides a
Model 1: HR (95% CI) 1.00 0.67 (0.47–0.94) 0.49 (0.34–0.70) ,0.0001 unique and thorough look at the topic
Model 2: HR (95% CI) 1.00 0.83 (0.58–1.18) 0.68 (0.47–0.97) 0.03 with data on the spectrum of possible
Quartile 3 breakfast intake frequencies as a dietary
n cases/n 112/328 43/182 69/329 behavior. This ability to examine the range
Incidence rate 17.9 11.5 9.7 of breakfast intake uniquely distinguishes it
Model 1: HR (95% CI) 1.00 0.70 (0.48–1.00) 0.65 (0.46–0.91) 0.008 from previous research on the topic and
Model 2: HR (95% CI) 1.00 0.79 (0.55–1.15) 0.79 (0.56–1.10) 0.13 better aligns the data with real-world be-
Quartile 4 (high) havior (8,15,16). Furthermore, the quality
n cases/n 78/254 38/175 77/442 of the overall dietary pattern is important
Incidence rate 15.9 10.4 8.0 for health (16), but this did not explain our
Model 1: HR (95% CI) 1.00 0.77 (0.51–1.16) 0.65 (0.46–0.92) 0.019 results, suggesting that the act of “breaking
Model 2: HR (95% CI) 1.00 0.93 (0.61–1.42) 0.83 (0.58–1.21) 0.33 the fast” may have important metabolic
Model 1 includes age, study center, race, sex, education, cigarette smoking, physical activity, alcohol con- health implications beyond the quality of
sumption, fast food restaurant use, dietary quality score, frequency of lunch/dinner and morning/afternoon/ the overall dietary pattern. The data from
evening snacks, and total energy intake. Model 2 includes model 1 + BMI (kg/m2) at baseline (year 7). Dietary the Health Professionals Study also sup-
quality score, an a priori score based on the reported intake of 46 food groups considered beneficial, neutral,
or adverse with a theoretical maximum of 132 and where a higher score indicates better diet quality; incidence
ports this assertion (13).
rate, incidence of metabolic syndrome per 1,000 person-years; metabolic syndrome, defined according to There are a number of plausible mech-
ATP III guidelines. anisms whereby eating breakfast may

3104 DIABETES CARE, VOLUME 36, OCTOBER 2013 care.diabetesjournals.org


improve acute and long-term factors salient
for metabolic risk. As summarized by Tim-
lin and Pereira (6), a spectrum of research
provides evidence that the act of eating
breakfast, as well as the content, plays im-
portant roles in factors related to appetite
and hormone, glucose, insulin, and lipid
metabolism. Indeed, the time of day and
frequency of eating, as well as content,
have important independent effects on en-
ergy intake, dietary content, and hormonal
response that are central to energy balance
and thus adiposity (30,31). Greater break-
fast intake frequency was strongly and in-
versely associated with long-term risk of
obesity in this study. This appears to be a
mechanism by which breakfast intake may
reduce risk for metabolic syndrome, type 2
diabetes, and hypertension. Since these es-
timates were not completely mediated by
adiposity in our statistical models, this sug-
gests that breakfast intake impacts other
avenues, likely hormonal glucose, insulin,
and lipid metabolism factors central to these
conditions (32).
A few studies with experimental de-
signs examined the effects of eating break-
fast or a larger portion of daily energy
intake in the morning on weight loss with
some suggestion of benefit, but with mixed
results possibly due to study design and
sample size (33,34). Other small experi-
mental studies have found that omitting
breakfast and the composition of breakfast
have effects on appetite and metabolic pa-
rameters that could impact long-term met-
abolic risk (35–37). Further trials have
demonstrated that the content of breakfast
is likely important for lower metabolic
risk as breakfast meals emphasizing low-
glycemic whole grains had beneficial effects
on appetite and metabolic parameters
throughout the day (36,38–40). The sensi-
tivity analysis we performed examining dif-
ferent breakfast-type foods aligns with this
point. Overall, the small base of prospective
and experimental research on breakfast in-
take suggests that it may have an indepen-
dent beneficial role in metabolic health.
Prospective population studies examining
breakfast habits and larger and longer ex-
perimental studies examining specific
mechanisms addressing both the act of
eating breakfast and the content would
provide a more definitive level of evidence
on breakfast intake and metabolic health.
They would also inform the level of empha-
sis that should be given to this dietary habit
in relevant dietary interventions and overall
dietary recommendations.
Our study has a number of strengths:
long-term, prospective study design, with
care.diabetesjournals.org
high rates of follow-up; standardized, valid,
and reliable measurements of dietary prac-
tices; extensive clinical measures and data
on covariates with which to explore con-
founders and mediators of the associations
under investigation; and the demographics
of the cohort, young adult black and white
men and women from four U.S. metropol-
itan areas who have been examined during
a period of life when substantial weight gain
occurs and metabolic complications de-
velop. Limitations include some level of
measurement error with the dietary assess-
ment, although this would most likely re-
sult in nondifferential misclassification
with respect to any of the outcomes and
likely underestimation of risk. The foods
explicitly consumed at breakfast were also
not assessed in CARDIA. Although a work-
ing scientific definition of breakfast has
been proposed in the time since CARDIA
began (6), there was no explicit definition
of what constituted a breakfast meal in
the CARDIA study. The self-report of other
lifestyle-related data may also result in
some misclassification and residual con-
founding in our models. Because of this,
it is possible that the association between
daily breakfast intake and significantly re-
duced metabolic risk is a proxy for an over-
all better diet and lifestyle.
In conclusion, young adults who re-
ported eating breakfast everyday had a
significantly lower risk of an array of
metabolic outcomes relative to their peers
who infrequently or never ate breakfast,
independent of adiposity and the overall
quality of the dietary pattern. Our study
and other burgeoning evidence (8–14) sug-
gest that the science is catching up to the
early nutritional beliefs related to the topic
and eating a daily breakfast meal is a dietary
habit that may be highly relevant for meta-
bolic health.
AcknowledgmentsdThe CARDIA Study is
funded by the CARDIA contract with the
National Heart, Lung, and Blood Institute
(N01-HC-48047 through N01-HC-48050 and
N01-HC-95095). D.S.L. was funded in part by a
career grant from the National Institute of Di-
abetes and Digestive and Kidney Diseases (K24-
DK-082730). A.O.O. and M.A.P. were funded
in part by a grant from the Bell Institute of
Health and Nutrition at General Mills. No other
potential conflicts of interest relevant to this
article were reported.
A.O.O. designed the analysis, performed
the statistical analysis, and wrote the manuscript.
D.R.J. and M.A.P. contributed to the design of
the analysis, interpreted data, and edited the
manuscript. L.M.S., L.V.H., and D.S.L. inter-
preted data and edited the manuscript. A.O.O.
DIABETES CARE, VOLUME 36, OCTOBER 2013
Odegaard and Associates
is the guarantor of this work and, as such, had
full access to all the data in the study and takes
responsibility for the integrity of the data and
the accuracy of the data analysis.
Parts of this study were presented in abstract
form at the 72nd Scientific Sessions of the
American Diabetes Association, Philadelphia,
Pennsylvania, 8–12 June 2012.
References
1. Lawrence F. Eat Your Heart Out: Why the
Food Business Is Bad for the Planet and Your
Health. Vol. 1. London, Penguin Books, 2008
2. Bernay E, Miller MC. Propaganda. Brooklyn,
NY, Ig Publishing, 2005
3. Tuttle WW, Wilson M, Daum K. Effect of
altered breakfast habits on physiologic re-
sponse. J Appl Physiol 1949;1:545–559
4. Ingoldsby ME. Be smartdeat a good
breakfast. Am J Nurs 1951;51:559–560
5. PHYSIOLOGIC results of breakfast hab-
its. Nutr Rev 1957;15:196–198
6. Timlin MT, Pereira MA. Breakfast fre-
quency and quality in the etiology of adult
obesity and chronic diseases. Nutr Rev
2007;65:268–281
7. U.S. Department of Agriculture; U.S. De-
partment of Health and Human Services.
Dietary Guidelines for Americans, 2010. 7th
Ed. Washington D.C., U.S. Government
Printing Office, 2011
8. Timlin MT, Pereira MA, Story M, Neumark-
Sztainer D. Breakfast eating and weight
change in a 5-year prospective analysis
of adolescents: Project EAT (Eating
Among Teens). Pediatrics 2008;121:
e638–e645
9. van der Heijden AA, Hu FB, Rimm EB, van
Dam RM. A prospective study of breakfast
consumption and weight gain among U.S.
men. Obesity (Silver Spring) 2007;15:2463–
2469
10. Bazzano LA, Song Y, Bubes V, Good CK,
Manson JE, Liu S. Dietary intake of whole
and refined grain breakfast cereals and
weight gain in men. Obes Res 2005;13:
1952–1960
11. Kochar J, Djoussé L, Gaziano JM. Breakfast
cereals and risk of type 2 diabetes in the
Physicians’ Health Study I. Obesity (Silver
Spring) 2007;15:3039–3044
12. Kochar J, Gaziano JM, Djoussé L. Break-
fast cereals and risk of hypertension in the
Physicians’ Health Study I. Clin Nutr 2012;
31:89–92
13. Mekary RA, Giovannucci E, Willett WC,
van Dam RM, Hu FB. Eating patterns and
type 2 diabetes risk in men: breakfast
omission, eating frequency, and snacking.
Am J Clin Nutr 2012;95:1182–1189
14. Smith KJ, Gall SL, McNaughton SA,
Blizzard L, Dwyer T, Venn AJ. Skipping
breakfast: longitudinal associations with
cardiometabolic risk factors in the
Childhood Determinants of Adult Health
Study. Am J Clin Nutr 2010;92:1316–
1325
3105
Breakfast and metabolic risk
15. So HK, Nelson EA, Li AM, et al. Breakfast
frequency inversely associated with BMI
and body fatness in Hong Kong Chinese
children aged 9-18 years. Br J Nutr 2011;
106:742–751
16. Sijtsma FP, Meyer KA, Steffen LM, et al.
Longitudinal trends in diet and effects of
sex, race, and education on dietary quality
score change: the Coronary Artery Risk
Development in Young Adults study. Am J
Clin Nutr 2012;95:580–586
17. Friedman GD, Cutter GR, Donahue RP,
et al. CARDIA: study design, recruitment,
and some characteristics of the examined
subjects. J Clin Epidemiol 1988;41:1105–
1116
18. Bild DE, Jacobs DR, Liu K, et al. Seven-year
trends in plasma low-density-lipoprotein-
cholesterol in young adults: the CAR-
DIA Study. Ann Epidemiol 1996;6:
235–245
19. Levine DA, Lewis CE, Williams OD, et al.
Geographic and demographic variability
in 20-year hypertension incidence: the
CARDIA study. Hypertension 2011;57:
39–47
20. Smith DE, Lewis CE, Caveny JL, Perkins
LL, Burke GL, Bild DE. Longitudinal
changes in adiposity associated with preg-
nancy. The CARDIA Study. Coronary Ar-
tery Risk Development in Young Adults
Study. JAMA 1994;271:1747–1751
21. McDonald A, Van Horn L, Slattery M,
et al. The CARDIA dietary history: devel-
opment, implementation, and evaluation.
J Am Diet Assoc 1991;91:1104–1112
22. Nettleton JA, Steffen LM, Ni H, Liu K,
Jacobs DR Jr. Dietary patterns and risk of
incident type 2 diabetes in the Multi-Ethnic
Study of Atherosclerosis (MESA). Diabetes
Care 2008;31:1777–1782
23. Nettleton JA, Schulze MB, Jiang R, Jenny NS,
Burke GL, Jacobs DR Jr. A priori-defined
dietary patterns and markers of
cardiovascular disease risk in the Multi-
3106 DIABETES CARE, VOLUME 36, OCTOBER 2013
Ethnic Study of Atherosclerosis (MESA).
Am J Clin Nutr 2008;88:185–194
24. Jacobs DR Jr, Sluik D, Rokling-Andersen
MH, Anderssen SA, Drevon CA. Associa-
tion of 1-y changes in diet pattern with
cardiovascular disease risk factors and adi-
pokines: results from the 1-y randomized
Oslo Diet and Exercise Study. Am J Clin
Nutr 2009;89:509–517
25. Liu K, Slattery M, Jacobs D Jr, et al.
A study of the reliability and comparative
validity of the cardia dietary history. Ethn
Dis 1994;4:15–27
26. Jacobs DR Jr, Hahn LP, Haskell WL, Pirie
P, Sidney S. Validity and reliability of a
short physical activity history: CARDIA
and the Minnesota Heart Health Program.
J Cardiopulm Rehabil 1989;9:448–459
27. National Cholesterol Education Program
(NCEP) Expert Panel on Detection, Eval-
uation, and Treatment of High Blood
Cholesterol in Adults (Adult Treatment
Panel III). Third Report of the National
Cholesterol Education Program (NCEP)
Expert Panel on Detection, Evaluation,
and Treatment of High Blood Cholesterol
in Adults (Adult Treatment Panel III) final
report. Circulation 2002;106:3143–3421
28. Breslow L, Breslow N. Health practices and
disability: some evidence from Alameda
County. Prev Med 1993;22:86–95
29. Liu S, Sesso HD, Manson JE, Willett WC,
Buring JE. Is intake of breakfast cereals
related to total and cause-specific mortality
in men? Am J Clin Nutr 2003;77:594–599
30. de Castro JM. The time of day of food
intake influences overall intake in hu-
mans. J Nutr 2004;134:104–111
31. Klok MD, Jakobsdottir S, Drent ML. The
role of leptin and ghrelin in the regulation
of food intake and body weight in hu-
mans: a review. Obes Rev 2007;8:21–34
32. Leroith D. Pathophysiology of the met-
abolic syndrome: implications for the
cardiometabolic risks associated with
type 2 diabetes. Am J Med Sci 2012;343:
13–16
33. Schlundt DG, Hill JO, Sbrocco T, Pope-
Cordle J, Sharp T. The role of breakfast in the
treatment of obesity: a randomized clinical
trial. Am J Clin Nutr 1992;55:645–651
34. Keim NL, Van Loan MD, Horn WF,
Barbieri TF, Mayclin PL. Weight loss is
greater with consumption of large morning
meals and fat-free mass is preserved with
large evening meals in women on a con-
trolled weight reduction regimen. J Nutr
1997;127:75–82
35. Farshchi HR, Taylor MA, Macdonald IA.
Deleterious effects of omitting breakfast
on insulin sensitivity and fasting lipid
profiles in healthy lean women. Am J Clin
Nutr 2005;81:388–396
36. Pal S, Lim S, Egger G. The effect of a low
glycaemic index breakfast on blood glu-
cose, insulin, lipid profiles, blood pressure,
body weight, body composition and sa-
tiety in obese and overweight individuals:
a pilot study. J Am Coll Nutr 2008;27:
387–393
37. Pereira MA, Erickson E, McKee P, et al.
Breakfast frequency and quality may affect
glycemia and appetite in adults and chil-
dren. J Nutr 2011;141:163–168
38. Barkoukis H, Marchetti CM, Nolan B,
Sistrun SN, Krishnan RK, Kirwan JP. A high
glycemic meal suppresses the postprandial
leptin response in normal healthy adults.
Ann Nutr Metab 2007;51:512–518
39. Nilsson AC, Ostman EM, Granfeldt Y,
Björck IM. Effect of cereal test breakfasts
differing in glycemic index and content of
indigestible carbohydrates on daylong
glucose tolerance in healthy subjects. Am J
Clin Nutr 2008;87:645–654
40. Isaksson H, Rakha A, Andersson R,
Fredriksson H, Olsson J, Aman P. Rye
kernel breakfast increases satiety in the
afternoon - an effect of food structure.
Nutr J 2011;10:31
care.diabetesjournals.org